Asta-Advances in Statistical Analysis最新文献

When estimating treatment effects in observational studies, propensity score analysis (PSA) is commonly used to reduce the arising bias that results from confounders interfering with causal inference. However, propensity score (PS) estimation is unstable if some confounders are densely measured and formed into high-dimensional data, which could eventually result in a biased estimate of the treatment effect. We propose two-stage analytic procedures to mitigate the high-dimensional problem: ridge PSA and functional PSA. In addition, conventional variance estimation of treatment effect estimates in the PSA methods tends to be biased, so we leverage the empirical bootstrap approach to develop a valid variance estimator. In the simulation study, we compare the bias and MSE of treatment effects estimated by ridge PSA and function PSA under the various confounding structures, including more densely measured confounders, and evaluate the performance of bootstrap variance estimators. The proposed methods are applied in the case study of police shootings.

In this paper, we define a penalized-distance likelihood function. This function is much more flexible than the available likelihood functions and can be used in many disciplines. Based on this function, we introduce a statistic for hypothesis testing and derive its asymptotic distribution. This statistic can be used to test a partial hypothesis in the parameter space for both non-sparse and sparse high-dimensional data. Relevant Bayesian analysis using the Markov chain Monte Carlo (MCMC) method will be discussed. Finally, we carry out a simulation study and apply our model to a real dataset.

Synthetic aperture radar (SAR) systems are highly efficient tools for addressing remote sensing challenges. They offer several advantages, such as operating independently of atmospheric conditions and producing high spatial resolution images. However, SAR images are often contaminated by a type of interference called speckle noise, which complicates their analysis and processing. Therefore, proposing statistical methods, such as regression models, that account for speckle behavior is an important step for users of SAR systems. In the work [ISPRS J. Photogramm. Remote Sens., 213, 1–13, 2024], the ({mathcal{G}^{0}_{I}}) regression model (short for (mathcal{R} {mathcal{G}^{0}_{I}})) was proposed as an interpretable tool to relate SAR intensity features to other physical properties. The authors employed maximum likelihood estimators (MLEs), known for their good asymptotic properties but prone to considerable bias in small and medium sample sizes. In this paper, we propose a matrix expression for the second-order bias of MLEs for (mathcal{R} {mathcal{G}^{0}_{I}}) parameters, based on the Cox and Snell method. This proposal is justified by the necessity of using small and moderate windows when processing SAR images, such as for classification and filtering purposes. We compare bias-corrected MLEs with their counterparts using both Monte Carlo experiments and an application to SAR data from a Brazilian region. Numerical evidence demonstrates the effectiveness of our proposal.

This paper considers a partially linear model in which the covariates of parametric part are measured with normal distributed errors. A newly robust corrected empirical likelihood procedure based on the corrected score function is proposed to attenuate the effects of measurement errors as well as outliers. What’s more, profit from the QR decomposition technique, the parametric and nonparametric components of the models can be estimated separately. The asymptotic properties of the proposed robust corrected empirical likelihood approach are established under some regularity conditions. Simulation studies are demonstrated to show that our proposed method performs well in finite samples. Boston housing price data are applied to illustrate the proposed estimation procedure.

We consider random coefficient INAR(1) processes with a strongly dependent latent random coefficient process. It is shown that, in spite of its conditional Markovian structure, the unconditional process exhibits long-range dependence. Short-term prediction and estimation of parameters involved in the prediction are considered. Asymptotic rates of convergence are derived.

The paper discusses the problem of estimating group heterogeneous fixed-effect panel data models under the assumption of fuzzy clustering, that is each unit belongs to all the clusters with a membership degree. To enhance spatial clustering, a spatio-temporal approach is considered. An iterative procedure, alternating panel data estimation and spatio-temporal clustering of the residuals, is proposed. The proposed method can be of relevance to researchers interested in using fuzzy group fixed-effect methods, but want to leverage spatial dimension for clustering units. Two empirical examples, the first on cigarette consumption in the US states and the second on non-life insurance demand in Italy, are presented to illustrate the performance of the proposed approach. The spatial fuzzy GFE model reveals important regional differences in both the US cigarette consumption and non-life insurance determinants in Italy. In the case of the US, we found a distinction in two main clusters, East and West. For the Italy provinces data, we find a distinction in North and South clusters. Regarding the regression results, for cigarette consumption data, different from the previous studies, we find that the smuggling effect is significant only in east regions, thus suggesting localised impacts of bootlegging. In the context of Italian non-life insurance demand, we find that while population density explains insurance consumption in northern provinces, the trust issues in the south explain the lower insurance demand.

This study investigates the spatial clustering and spillover effects of COVID-19 vaccine uptake in Romania, focusing on the municipality-level distribution of vaccine acceptance and hesitancy while considering the factors that influence it. The research uses the Spatial Durbin Error Model (SDEM) and identifies spatial clusterization, as well as significant contagion and diffusion processes in the vaccination behaviour conditioned by socioeconomic factors, labour market characteristics, social and religious attitudes, urban, and health indicators. We find evidence in favour of spatial spillover effects of the poverty rate, opposition to same-sex marriage, COVID-19 infection rate, peri-urban towns, and denser cities. Our findings contribute to the literature of the spatial distribution and determinants of vaccine uptake and carry practical implications for policy makers offering evidence-based insights that can inform targeted strategies and interventions to enhance vaccine acceptance and address hesitancy in specific locations.

In this paper, we propose an integrated modified harmonic mean estimator (IHME) for nested and non-nested model selection problems in spatial panel data models with entity and time fixed effects. We formulate the IHME based on the integrated likelihood functions obtained by analytically integrating out the high-dimensional entity and time fixed effects from the complete likelihood functions. To investigate the finite sample properties of the IHME, we design a comprehensive simulation study that allows for both nested and non-nested model selection exercises in some popular spatial panel data models. Our simulation results show that the IHME has excellent finite sample performance and outperforms some competing estimators in terms of precision. We provide an empirical application on the US house price changes to show the usefulness of the proposed IHME in a model selection exercise.

The primary purpose of this paper is to assess households’ burden due to out-of-pocket healthcare expenditures. These payments are modeled on a representative sample of 25668 Italian households as the fraction of out-of-pocket healthcare expenditures over the households’ capacity to pay. For this purpose, we propose extending the analysis of the so-called catastrophic payments by looking at the entire distribution of this ratio. We introduce a novel finite mixture regression able to capture different levels of heterogeneity in the data. By using such a model specification, the fairness of the Italian National Health Service and its determinants are investigated.

In this work, the distributional properties of the goodness-of-fit term in likelihood-based information criteria are explored. These properties are then leveraged to construct a novel goodness-of-fit test for normal linear regression models that relies on a nonparametric bootstrap. Several simulation studies are performed to investigate the properties and efficacy of the developed procedure, with these studies demonstrating that the bootstrap test offers distinct advantages as compared to other methods of assessing the goodness-of-fit of a normal linear regression model. Our inferential technique can be employed using the DBModelSelect R package, available freely via the Comprehensive R Archive Network.